Skip to main content

Does Foreign Environmental Policy Influence Domestic Innovation? Evidence from the Wind Industry


This paper analyses the relative influence of domestic and foreign demand-pull policies in wind power across OECD countries on the rate of innovation in this technology. We use annual wind power generation to capture the stringency of the portfolio of demand-pull policies in place (e.g., guaranteed tariffs, investment and production tax credits), and patent data as an indicator of innovation activity. We find that wind technology improvements respond positively to policies both home and abroad, but the marginal effect of domestic policies is 12 times greater. The influence of foreign polices is reduced by barriers to technology diffusion, in particular lax intellectual property rights. Reducing such barriers therefore constitutes a powerful policy leverage for boosting environmental innovation globally.

This is a preview of subscription content, access via your institution.

Fig. 1


  1. 1.

    See, for example, President Obama’s speech at the Massachussets Institute of Technology, October 23rd, 2010: “The world is now engaged in a peaceful competition to determine the technologies that will power the 21st century (\(\ldots \)). The nation that wins this competition will be the nation that leads the global economy. And I want America to be that nation.” Similar statements were made by political leaders in many countries.

  2. 2.

    The restriction to OECD countries stems from the unavailability of data on public R&D expenditures for non-OECD countries. For consistency we estimate the diffusion equation on the same sample.

  3. 3.

    These are Sinovel, Goldwind, Dongfang, United Power (China) and Suzlon (India). The other companies in the top 10 in 2011 were Vestas (Denmark), GE (USA), Enercon and Siemens (Germany) and Gamesa (Spain).

  4. 4.

    Peters et al. (2012) look at 15 OECD countries across 1978–2005, while we consider 28 OECD countries between 1991 and 2008. Furthermore, they distinguish between continental and intercontinental demand, while we aggregate both into foreign demand.

  5. 5.

    An overview of the measures adopted by every country, including the timing of their adoption, is available from the IEA/IRENA Global Renewable Energy Policies and Measures database, available at (last accessed 1 July 2013).

  6. 6.

    Most recently in July 2012, Japan introduced a feed-in tariffs scheme, obliging incumbent power companies to buy the output from solar, wind, geothermal, small hydro and some biogas and biomass-fueled plants at premium prices.

  7. 7.

    Johnstone et al. (2010) is a notable exception. However, except for feed-in tariffs, they only measure the strictness of the different policy instruments by a binary variable, with one indicating that the particular instrument is in place.

  8. 8.

    We use the October 2012 version of the PATSTAT database, and it takes on average 3 years for a patent to be granted. Note that our results are robust to changing the end year to 2007 or 2009.

  9. 9.

    Note that Least Developed Countries are not present in our dataset, for two related reasons: their patenting activity is extremely limited, and available statistics are not reliable.

  10. 10.

    In addition, we randomly sampled 100 patents from the F03D category and checked their relevancy based on title and abstract. We found only three irrelevant patents, which lead us to believe the patent classification is accurate for wind power technologies.

  11. 11.

    For 1.4 % of the patent applications included in our dataset, the inventor’s country of residence is not available. When this information is missing, we simply assume that the inventor’s country corresponds to the first patent office in which protection was taken (i.e. the priority office).

  12. 12.

    Patents with multiple inventors are counted fractionally. For example, if two inventor countries are involved in an invention, then each country is counted as one half.

  13. 13.

    Our results are robust to using all filed patent applications, however.

  14. 14.

    Family size (the number of countries in which a patent is filed) is another way of assessing the value of a patent. But we think it is better to use patent citations in this particular paper, given the questions addressed. The problem is that family size is not only a value indicator; it also captures the degree of internationalization of the invention, which is, roughly speaking, the central topic of the paper. This suggests using family size as a dependent variable, not as a weight when constructing an independent variable. To a certain extent, this is what we do in Sect. 5 where the dependent variable is the bilateral flow of patent between countries.

  15. 15.

    In fact, about 75 % of inventions are patented in only one country.

  16. 16.

    Available at

  17. 17.

    An alternative option is to estimate installed capacity in non-OECD countries by running a regression of energy capacities on energy production using the data from OECD countries and then use this model to make out-of-the-sample predictions for capacities in non-OECD countries based on their production. We implemented this method, which gives qualitatively similar results. It however has two weaknesses: first, the relationship between capacity and production might differ between OECD and non-OECD countries; second, using predicted values would lead us to underestimate standard errors in the subsequent regression analysis.

  18. 18.

    Excluding patents filed at the European Patent Office, the figure is 42 %.

  19. 19.

    We downloaded data on wind-power generating sets (product code HS 850231).

  20. 20.

    More precisely we count all citations made by patents applied for up to five years after the publication of each patent. Note that PATSTAT includes citation information from 98 patent offices.

  21. 21.

    As mentioned above, inventions patented in several countries are only counted once in order to avoid double-counting. We restrict patent data to private inventions only to avoid potential endogeneity problems.

  22. 22.

    In other words, we include a full set of country dummies in the estimation. Another way to deal with fixed effects would be to use the conditional maximum likelihood estimator introduced by Hausman et al. (1984) and available in STATA as the xtnbreg command. However this model is known to imperfectly control for fixed effects (Allison and Waterman 2002; Greene 2007). Another issue is that xtnbreg does not report any robust or clustered standard errors and the small size of our sample has not allowed us to compute bootstrapped standard errors.

  23. 23.

    Recall however that differences in sample size and in the way explanatory variables are constructed make it difficult to accurately compare the results between the two papers (see note 4 above).

  24. 24.

    See (last accessed 24 May 2013). This corresponds to an annual production of 37.4 GWh.

  25. 25.

    Recall that the value of patents is heterogeneous. Therefore, these figures describe the effect of policies on the average invention.

  26. 26.

    Cognitive limitations of innovators could provide an alternative explanation: they simply ignore the installations of new wind farms in certain foreign countries, which lead them to infer that the demand is actually zero. Note this interpretation rests on a bounded rationality assumption: A rational decision maker under uncertainty will consider the expected size of the market derived from a prior subjective probability distribution.

  27. 27.

    Note that multiplying the effect by 27 only yields the aggregate effect in OECD countries. We cannot calculate the effect of innovation in non-OECD countries as they are not included in the estimation sample.

  28. 28.

    The size of this effect centrally depends on the value of the discount rate \(\delta \). With \(\delta =0.15\), the additional long term impact of both demand-pull policies and public R&D through increased knowledge stock is about one half of the short-term impact.

  29. 29.

    Peri (2005) shows that only 12 % of the knowledge created in a country spills over to foreign countries.

  30. 30.

    This is the reason why we did not use R&D expenditures in marine energy as an instrument. The technologies used for marine and wind energy production have some similarities. We considered adding public R&D expenditures in biomass and geothermal energy as additional instruments but none of them turned up significant. However, the results are completely robust to including them.


  1. Aghion P, Dechezleprêtre A, Hemous D, Martin R, Van Reenen J (2012) Carbon taxes, path dependency and directed technical change: evidence from the auto industry. NBER working paper 18596, National Bureau of Economic Research, Inc

  2. Allison PD, Waterman RP (2002) Fixed-effects negative binomial regression models. Sociol Methodol 32:247–265

    Article  Google Scholar 

  3. Brunnermeier SB, Cohen MA (2003) Determinants of environmental innovation in US manufacturing industries. J Environ Econ Manag 45:278–293

    Article  Google Scholar 

  4. Buchanan B, Keefe P (2010) Environmentally sound technology (EST): an analysis of the classification of EST energy sectors in the IPC; and of UK EST innovation through the IPC. Paper presented at the patent statistics for decision makers conference, 17–18 November 2010, Vienna, Austria

  5. COMTRADE (2012) United Nations Commodity Trade Statistics Database. Available at (last accessed 1 July 2013)

  6. Crabb J, Johnson D (2010) Fueling innovation: the impact of oil prices and CAFE standards on energy-efficient automotive technology. Energy J 31(1):199–216

    Article  Google Scholar 

  7. Dechezleprêtre A, Glachant M, Ménière Y (2013) What drives the international transfer of climate change mitigation technologies? Empirical evidence from patent data. Environ Resour Econ 54(2):161–178

    Article  Google Scholar 

  8. Dechezleprêtre A, Glachant M, Johnstone N, Haščič I, Ménière Y (2011) Invention and transfer of climate change mitigation technologies: a global analysis. Rev Environ Econ Policy 5(1):109–130

    Article  Google Scholar 

  9. Dekker T, Vollebergh HRJ, De Vries FP, Withagen C (2009) Inciting protocols. J Env Econ Manag 64(1):45–67

    Article  Google Scholar 

  10. Eaton J, Kortum S (1996) Trade in ideas: patenting and productivity in the OECD. J Int Econ 40(3–4):251–278

    Article  Google Scholar 

  11. Eaton J, Kortum S (1999) International technology diffusion: theory and measurement. Int Econ Rev 40(3):537–570

    Article  Google Scholar 

  12. ECOOM-EUROSTAT-EPO PATSTAT Person Augmented Table (EEE-PPAT) database (2012). Last accessed 1 July 2013

  13. Greene W (2004a) Fixed effects and bias due to the incidental parameters problem in the tobit model. Econom Rev 23(2):125–147

    Article  Google Scholar 

  14. Greene W (2004b) The behaviour of the maximum likelihood estimator of limited dependent variable models in the presence of fixed effects. Econom J 7(1):98–119

    Article  Google Scholar 

  15. Greene W (2007) Fixed and random effects models for count data. Working paper Department of Economics, Stern School of Business, New York University, New York

    Google Scholar 

  16. Griliches Z (1990) Patent statistics as economic indicators: a survey. J Econ Lit 28(4):1661–1707

    Google Scholar 

  17. Hall BH, Jaffe A, Trajtenberg M (2005) Market value and patent citations. Rand J Econ 36:16–38

    Google Scholar 

  18. Hausman J, Hall BH, Griliches Z (1984) Econometric models for count data with application to the patents-R &D relationship. Econometrica 52:909–938

    Google Scholar 

  19. Helfgott S (1993) Patent filing costs around the world. J Pat Trademark Off Soc 75:567–580

    Google Scholar 

  20. Hendry C, Harborne P (2011) Changing the view of wind power development: more than bricolage. Res Policy 40(5):778–789

    Article  Google Scholar 

  21. IEA (2003) Renewables for power generation. International Energy Agency, Paris

    Google Scholar 

  22. IEA (2005) Projected costs of electricity generation. International Energy Agency, Paris

    Google Scholar 

  23. IEA (2009) Wind technology roadmap. International Energy Agency, Paris

    Google Scholar 

  24. IMF (2012) Annual report on exchange arrangements and exchange restrictions. International Monetary Fund, Washington, DC

  25. Jaffe AB, Palmer K (1997) Environmental regulation and innovation: a panel data study. Rev Econ Stat 79(4):610–619

    Article  Google Scholar 

  26. Johnstone N, Haščič I, Popp P (2010) Renewable energy policies and technological innovation: evidence based on patent counts. Environ Resour Econ 45(1):133–155

    Article  Google Scholar 

  27. Keller W (2004) International technology diffusion. J Econ Lit 42(3):752–782

    Article  Google Scholar 

  28. Lanjouw JO, Mody A (1996) Innovation and the international diffusion of environmentally responsive technology. Res Policy 25:549–571

    Article  Google Scholar 

  29. Lanjouw J, Pakes A, Putnam J (1998) How to count patents and value intellectual property: uses of patent renewal and application data. J Ind Econ 46(4):405–432

    Article  Google Scholar 

  30. Nemet G (2009) Demand-pull technology-push and government-led incentives for non-incremental technical change. Res Policy 38(5):700–709

    Article  Google Scholar 

  31. Neuhoff K (2005) Large-scale deployment of renewables for electricity generation. Oxf Rev Econ Policy 21(1):88–110

    Article  Google Scholar 

  32. Newell RG, Jaffe AB, Stavins RN (1999) The induced innovation hypothesis and energy-saving technological change. Q J Econ 114:941–975

    Article  Google Scholar 

  33. OECD (2009) OECD patent statistics manual. Paris

  34. Park WG, Lippoldt DC (2008) Technology transfer and the economic implications of the strengthening of intellectual property rights in developing countries. OECD Trade Policy working papers 62, OECD Trade Directorate, Paris

  35. PATSTAT (2012) Worldwide patent statistical database. European Patent Office. October 2012

  36. Peri G (2005) Determinants of knowledge flows and their effect on innovation. Rev Econ Stat 87(2):308–322

    Article  Google Scholar 

  37. Peters M, Schneider M, Griesshaber T, Hoffmann VH (2012) The impact of technology-push and demand-pull policies on technical change: does the locus of policies matter? Res Policy 41(8):1296–1308

    Article  Google Scholar 

  38. Popp D (2002) Induced innovation and energy prices. Am Econ Rev 92(1):160–180

    Article  Google Scholar 

  39. Popp D (2006) International innovation and diffusion of air pollution control technologies: the effects of NOX and SO2 regulation in the US Japan and Germany. J Environ Econ Manag 51:46–71

    Article  Google Scholar 

  40. Popp D, Hafner T, Johnstone N (2011) Policy vs. consumer pressure: innovation and diffusion of alternative bleaching technologies in the pulp industry. Res Policy 40(9):1253–1268

    Article  Google Scholar 

  41. Schlenker W, Walker WR (2011) Airports, air pollution, and contemporaneous health. NBER Working Papers 17684, National Bureau of Economic Research, Inc.

  42. Verdolini E, Galeotti M (2011) At home and abroad: an empirical analysis of innovation and diffusion in energy technologies. J Env Econ Manag 61:119–134

    Article  Google Scholar 

  43. WIPO (2010) World Intellectual Property Indicators 2010

  44. Wooldridge (2002) Econometric analysis of cross section and panel data. MIT Press, Cambridge, MA

Download references


The authors thank David Popp and three anonymous reviewers for their helpful comments and suggestions. We also thank many conference and seminar participants in Toulouse, Paris, Prague, Mannheim, and London. Financial support by the French Council for Energy (CFE) is gratefully acknowledged.

Author information



Corresponding author

Correspondence to Matthieu Glachant.

Appendix: Robustness Checks

Appendix: Robustness Checks

A number of robustness tests were conducted and the main ones are reported below.

Poisson Estimator

As an alternative to the negative binomial specification, we reestimate equations presented in Table 5 using conditional maximum likelihood Poisson with fixed effects (Hausman et al. 1984). Results are closely similar (see Table 6, columns 1 and 2). Most importantly, the foreign installations variable remains positive and highly statistically significant in both specifications.

Accounting for the Potential Endogeneity of Public R&D Expenditures

The variable \(\ln rd_{i,t-1} \) may pose a simultaneity problem as domestic R&D expenditures are inputs of the innovation process. Although we exclude public patents from the dependent variable, public R&D expenditures as reported by the IEA include tax credits on private R&D expenditures, which may give rise to endogeneity bias. We address this issue by using an instrumental variables approach. R&D public expenditures in solar and hydro power in the same country and year are used as instruments. R&D expenditures in these domains present the necessary properties. First, they do not directly influence the number of wind patents as they differ from wind energy from a technological point of view.Footnote 30 Second, they are positively correlated with \(rd_{i,t}\) as there is arguably a degree of jointness in the policy decisions to support R&D in specific renewable technology fields. Since our estimation uses a Poisson model we adopt the control-function approach suggested by Wooldridge (2002). In the first stage we regress \(\ln rd_{i,t}\) on the instrumental variables and the exogenous variables in Eq. (2) using a log-linear estimation, and in the second stage we include the residual of the first stage estimation as an additional regressor in Eq. (2) (see Schlenker and Walker (2011), for a recent application).

The first stage estimation together with the usual statistics are presented in Table 7. Column 1 shows the results from the unweighted specification and column 2 shows the results from the weighted specification. The coefficient of the excluded instruments is statistically significant and positive. The cluster-robust F-statistics of joint significance of the two instruments are 5.18 and 5.28 (\(p\) value of 0.01 in both cases) respectively. This suggests the instruments do a resonably good job. Results from the second stage equation are shown in columns 3 and 4 of Table 6. The results remain similar to our baseline estimates but the point estimates for \(\ln rd_{i,t} \) increase in both specifications. However, the coefficient on the residuals from the first stage equation are not significantly different from 0, suggesting that the hypothesis that \(\ln rd_{i,t-1} \) is exogenous cannot be rejected. Overall, results from these tests suggest that our baseline estimates should be viewed as a lower bound estimate of the impact of public R&D expenditures. Importantly, results concerning domestic and foreign demand are robust to instrumenting public R&D.

Table 7 First stage equations

Other Tests

As is commonly the case with patent data, the distribution of patents across countries is highly heterogeneous, with a few countries accounting for a large share of innovations. For this reason, it is necessary to check that our results are not driven by outliers. Columns 5 and 6 of Table 6 reports the results obtained when we drop Japan, by far the top inventor in our sample with 35 % of the total patented inventions. Our findings are robust, although the point estimate obtained on domestic installations decreases.

Finally, applying alternative discount rate values which are used to calculate the knowledge stocks—specifically, 10 and 20 %—made no difference to the results (robustness test results not shown).

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Dechezleprêtre, A., Glachant, M. Does Foreign Environmental Policy Influence Domestic Innovation? Evidence from the Wind Industry. Environ Resource Econ 58, 391–413 (2014).

Download citation


  • Innovation
  • International technology diffusion
  • Renewable energy policy
  • Wind power

Jel Classification

  • O31
  • Q42
  • Q55